A recent article published in the scientific journal Construction and Building Materials presents a comprehensive study on the factors that control susceptibility to landslides in a mountainous area of the Ecuadorian Andes, incorporating geomorphological analyses, lithological variables, and advanced machine learning models.
The research addresses a critical issue for the country: the recurrence of mass movements associated with steep slopes, intense rainfall, and geological complexity, which impact road infrastructure, human settlements, and strategic ecosystems. Given this context, the study proposes a quantitative approach that improves the identification of vulnerable areas and strengthens territorial planning.
Construction of the technical basis for the study
The work began with the development of a multi-temporal inventory of landslides, validated through detailed geospatial analysis. Based on this information, a database was structured with conditioning variables such as slope, terrain orientation, curvature, elevation, lithology, distance to drains, and proximity to roads.
The integration of these factors allowed for a systematic analysis of the relationship between the physical characteristics of the territory and the occurrence of mass movements, establishing spatial patterns and prioritizing variables with the greatest influence.
Application of machine learning models
One of the most relevant contributions of the study is the comparative application of machine learning algorithms, including Random Forest and Support Vector Machine (SVM), with the aim of generating high-precision susceptibility maps.
The performance of the models was evaluated using robust statistical metrics such as the Area Under the ROC Curve (AUC), which allowed their predictive capacity to be validated. The results showed that models based on assemblages perform better in contexts of high geological heterogeneity, significantly improving the identification of critical areas.
Determining factors in slope instability
The analysis confirmed that susceptibility to landslides does not depend on a single factor, but rather on the complex interaction between topographic, geological, and hydrological variables. Among the factors with the greatest predictive weight are slope gradient, lithology type, and proximity to drainage systems.
These findings reinforce the importance of incorporating integrated approaches that combine geomorphology, geology, and artificial intelligence to understand and model instability processes in mountainous regions.
Territorial impact and contribution to sustainability
As a result, the study generated a susceptibility map classified into different threat levels, constituting a key tool for land use planning, resilient infrastructure design, and disaster risk management in Andean territories.
